From Idea to Invention

Textile Innovation at Otherlab

At Otherlab we are developing textiles that do something delightfully unexpected: they move and change shape, increasing their thickness and insulation in response to the cold. The change is a physical response, occurring without input from the wearer or any control system. We are trying to make clothing that is so thermally comfortable that the heating and cooling of buildings is impacted.

Photograph of a single layer thermally adaptive textile prototype.

As with many ideas and advances, what appears to be new is a combination of the old. About 15 years ago, one of Otherlab’s founders, Saul Griffith, was in a MEMS class at MIT, working on a cantilever bimorph. A bimorph contains two dissimilar materials, and as the materials change lengths in response to a temperature change, the bimorph bends. Around the same time, Saul was prototyping a haptic feedback system, using resistance heating to control shape memory wires that he had carefully stitched into a knit glove. We shared an office and I remember him in our basement lab shaking the glove off his burning hand when the controls for the glove didn’t work as planned. Saul already had a rich background in textiles when he came across the bimorph structure. Although it was in a completely different domain, he transported the bimorph concept from one space into another and saw a way to actuate a textile structure. This insight seeded a vision (what: a variable textile) and a mechanism (how: a bimorph) for an adaptive textile that wouldn’t really be worked on or worked out for more than a decade.

Saul revived his old idea at Otherlab, and a small team put together a pre-proposal for ARPA-E, the advanced research arm of the Department of Energy. The short paper mathematically sketched out the performance limits for a bimorph. They made some sensible assumptions, such as using commodity materials with standard thicknesses, and they calculated the limits for a bimorph bending over a narrow 10 ºC range. In a textile the calculated change could be enough to impact human thermal comfort, enough to enable people to use less energy heating and cooling, enough to potentially save more than 1% of domestic energy and related emissions.

An illustration of a simple bimorph beam structure. The dashed lines illustrate the beam in a bent position at a different temperature than the beam’s initial position, shown with solid lines. Key dimensions and material parameters are labeled.

I inherited the idea shortly after I joined Otherlab, just as most of the early developers of the idea were leaving the lab. It became my responsibility to transform an idea that had been shown to be conceptually feasible into an idea that might be practically attainable. I promptly came up with reasons the early ideas for implementation wouldn’t approach the ideal performance limits that had been calculated: bimorphs would impede each other’s motion or neighboring bimorphs wouldn’t move in a coordinated way so that most of their movement would be unproductive. I was able to poke some pretty large holes in a lot of our early ideas. I came up with some new ideas, only to find they were no better, and I spent a lot of time sketching out structures that were hopelessly un-manufacturable, structures that worked locally but failed in the aggregate, and structures that were impractical in three dimensions but initially looked wonderful in the two dimensional plane of a sheet of paper. I closed a lot of doors and opened none.

I spent about 80 hours examining our ideas, staring at the basic bimorph equations, discussing the problems I saw with anyone that would listen around the lab, and I came out the other side with a proposal. The math said our vision was possible, but most of my progress was in culling ideas that had no chance, leaving me with a handful of ideas that still looked like long shots. But we believed the math and I told myself that long shots work out sometimes; ARPA-E agreed and awarded us with up to two years of funding. It was affirming, but I kept thinking about the challenges facing the project as we wrote and negotiated the formal project milestones.

One of the problems we had identified early on was order. A typical shirt has many millions of fibers in it, constituting many thousands of individual yarns that are woven or knit and then cut and sewn to produce the garment. The structural order in a textile spans more than 3 orders of magnitude of size as we move from fiber through to finished textile. Our research has to consider and coordinate thermal response at each level if we are to produce the envisioned textile.

Researchers at the US Army Soldier Center in Natick, MA, had done some work modeling and then producing fiber battings made of bicomponent bimorph fibers. The approach was simple, maybe even elegant, but the disorder in the random batting seemed like it would produce structures that underwent only a moderate amount of dimensional change in response to temperature change; their experimental work demonstrated a change in insulation that wasn’t even a tenth of what ARPA-E was targeting. Our analysis had told us that bimorphs could achieve the change we were looking for, but the math assumed an organized structure with bimorphs behaving cooperatively. Optimal performance required order across all scales, from fiber through to fabric. Randomness in the textile structure would not be acceptable.

We decided to move away from individual bimorph fibers and began to focus on laminated sheet structures, which offered us lateral structure or order within a sheet, easy handling for manufacturing, and a number of routes to ordering individual bimorph sheets so that we might make multilayer structures. We solved a problem at the fiber level by moving to sheets or ribbons where rotation and randomness were reduced. While an individual bimorph sheet would transition from a flat layer into a wavy structure, multiple bimorph sheets could easily be stacked on each other to create a structure with small air-trapping pockets. As each bimorph sheet changes shape, the size of the air-pockets between layers change, augmenting overall thickness and the insulation of the multilayer structure. Sheets or thin film membranes are already used in some garments, perhaps most notably in waterproof outerwear that uses Gore-Tex, but our move to sheet structures introduced a new set of challenges.

One of the problems I had overlooked early on was molecular order. This was particularly embarrassing because I am trained as a chemist. And at Otherlab I am the one person trained as a chemist, so there is no one else who can take the blame. Of course, that means I blame Saul. The observed properties of a material depend on the structure of the material, right down to the molecular level. This is true for the polymer fibers and sheets that we were looking at, and it is also true that I had overlooked the importance of molecular order in relation to material processing. Stretching a polymer fiber or thin film can bring about polymer chain alignment, leading to interesting and desirable anisotropic properties including high strength and stiffness. When highly stretched, some polymers have very small, even negative thermal expansion coefficients, which was of obvious interest to us as we sought out pairs of materials with different thermal expansion properties. Unfortunately, the obvious way to connect these materials — thermal bonding — would upset the molecular order that made their properties interesting in the first place. Shara Maikranz, who had started working with me on the project, prototyped some good-looking structures using adhesives, but adhesives added mass and thickness, both of which reduced performance. Furthermore, sheet structures would also require invention around perforation or slitting of the bimorph sheet so that it could appropriately breathe, stretch, and flex.

An early prototype bimorph sheet structure in a lofted state. A change in temperature causes the sheet to flatten, reducing its effective thickness.

The problems associated with bimorph fibers had led us to consider bimorph sheets, and what we learned from those sheet structures we translated into textile structures. Just as the sheet structure had solved problems with the fiber structure, the textile structure promised to solve problems with the sheets, namely manufacturability, breathability, and flexibility. It was only through solving the fiber level problem in a simplified sheet structure that I was able to see viable adaptive structures clearly enough to imagine how they might be reproduced in a textile structure, a structure where the bimorph is not built into a single bicomponent fiber but is built into a textile made with two different fibers. It had been about a year since I first started thinking about adaptive textiles. Shara and I reviewed a hundred resumes, eventually interviewing and hiring Jean Chang, who helped transform our new idea into a first textile prototype. We had a small team and started building up our prototyping, testing, and analysis capabilities.

After investigating and walking away from idea after idea, we had come to a new approach that captured the insights of the best of our old ideas and retained the essential portions of the earliest vision, an adaptive textile actuated with bimorph structures. We sought to produce these adaptive materials in woven and knit structures and hired Leah Bryson to do textiles design and production. Through the combined efforts of the four of us we were able to produce a first woven adaptive textile.

Otherlab’s first woven adaptive textile passively responding to temperature. The structure of the textile pairs two materials with different thermal expansion characteristics, resulting in the unique behavior.

As with many ideas and advances, what appears to be new is the product of more than just a moment where two old ideas collide. It has been 15 years, and the vision and mechanism remain the same for our adaptive textiles, but all of the specifics relating to implementation are different. Multiple rounds of additional innovation and insight have been required. Some of the techniques we are using now we didn’t know about 15 years ago. What we are doing now only came about through an extended period of unearthing ill-formed ideas, extracting from them what we could, and moving on. Most of our ideas and moves were wrong. The right thing we did, again and again, was learn from what was wrong. And we’re still learning.

Our goal is to make apparel with previously unknown thermal comfort, indoors and outside. The wearer’s experience should be almost magical, subtle enough that the new comfort feels completely natural, but at the same time the thickness changing enough to be seen and discerned with the eye. We are now in a period of rapid iteration, experimentally exploring the relationships between materials and structure, from the fiber level all the way up to the multiple layers of a finished textile product. Our conversations with individuals from leading textiles companies have been promising, but we are looking to have many more around advanced garment design, textiles production, and translating our work from the lab to the market. If this journey or destination excites you, let us know. We’d love to talk. We’ve sifted through a hundred ideas and are looking forward to sifting through a hundred more.